Medical editor Malachi Sheahan III, MD, reflects on the release of the CREST-2 results and considers what they may mean for clinical practice.
As you likely know, the long-awaited CREST-2 trial results were recently released. While panic and jubilation erupt among different tribes of our fractured health care system, the findings demand a measured scientific look. Of course, you didn’t come to me for that, so I’ll point you in the direction of the official Society for Vascular Surgery (SVS) response published by AbuRahma, Schermerhorn, and Calligaro in the December Journal of Vascular Surgery (JVS). What follows is my (very biased) opinion, FAQ style.
What was the design of CREST-2?
Two parallel, observer-blinded randomized controlled trials (RCTs) in patients ≥35 years with ≥70% asymptomatic carotid stenosis. One compared intensive medical management (IMM) alone vs. IMM + carotid artery stent (CAS), the other IMM vs. IMM + carotid endarterectomy (CEA). Primary endpoints: stroke or death within 44 days, plus ipsilateral ischemic stroke up to four years.
What were the main outcomes for CAS?
At four years risk: 6.0% in IMM vs. 2.8% in CAS + IMM (P = .02). Periprocedural risk: 1.3% with CAS. Number needed to treat was 31.
What were the main outcomes for CEA?
Four-year composite: 5.3% IMM vs. 3.7% CEA + IMM (P = .24). Periprocedural: IMM had 3 strokes; CEA had 9 strokes.
How did CAS compare to CEA?
Both had lower risk of stroke at four years than IMM but only CAS reached significance. Both had low rates of disabling stroke rates.
What was considered intensive medical therapy?
BP <130 mmHg, LDL <70 mg/dL, lifestyle optimization.
How close was CEA to achieving statistically significant improvement over IMM?
A tipping point analysis showed that six more strokes in the IMM group would have given this result. Some back-of-the-envelope math suggests that if 29 strokes occurred in the 600 medical group patients at four years, stroke reduction with CEA would have achieved significance. Interestingly, there were 28 strokes in the 600 medical group patients in the CAS trial.
Did CAS outperform CEA?
No. These were two separate trials with different inclusion criteria. A direct comparison between CAS and CEA cannot be made. As I pointed out, it appears the benefit of CEA would have achieved significance had it been compared to the same medical management group as CAS.
Over the past 15 years, seven of the eight randomized controlled trials directly comparing these two modalities demonstrated a persistently increased incidence of 30-day stroke with CAS over CEA (including CREST-1 by the way). A finding reinforced by real world VQI data.
Anyone claiming that CAS outperformed CEA in CREST-2 would be using the same transitive property logic fail that led my son to claim his soccer team was the best since they beat the eventual champs (ignoring their 0-12 record against the rest of the division).
With these results so close I am sure the conclusions drawn by other specialties have been tempered?
My sweet summer child. From the New England Journal of Medicine: “We can conclude that there is no longer a role for routine carotid endarterectomy in persons with asymptomatic stenosis.”
Here is a just a sample of the headlines seen on page 1 of a Google search:
– A New Asymptomatic Carotid Stenosis Paradigm?
– Long-Awaited CREST-2 Results Bolster Stents for Asymptomatic Carotid Stenosis
– CREST-2 Shows Reduced Stroke Risk With Carotid Artery Stenting Plus Intensive Medical Management; No Added Benefit Seen With CEA
– Carotid Stenting Makes a Strong Showing in the CREST-2 Registry
– Asymptomatic Blocked Carotids Don’t Need Surgery, Large Trial Shows
Social media starts here: “Verdict is out!!!!! #CREST2 RESULTS FAVORING CAS!!!” and devolves quickly.
Wait, the periprocedural stroke risk was 1.3% with CAS? Is that typical?
No, it is historic. This is by far the lowest reported in any randomized carotid revascularization trial with independent outcome adjudication. Stent enthusiasts will credit improvements in device technology, imaging, and medical management. Skeptics will note that the credentialing process was incredibly rigorous with about half of the prospective interventionalists rejected. Conversely, about 90% of the surgeons were accepted.
Life in the medical arms of this trial seems nice. How can I move there?
I know right? Free meds! Lifestyle coaches! I’ve got patients trying to light up Marlboros in the PACU while their family is trafficking Popeyes to their bedside. This trial shows us the promise of medical therapy, but not necessarily our reality. And even in this ideal scenario, the results may not be as dramatic as they appear. As the official SVS response points out, the annual stroke risk with medical therapy in the (now ancient) Asymptomatic Carotid Artery Study was 2% and decreased to 1% over time. In CREST-2, medical management yielded an annual stroke rate of 1.7% in the CAS trial and 1.5% in the CEA trial.
Enrollment began in 2014, before the widespread use of TCAR. Since TCAR has never been studied in a randomized controlled manner, this remains an open, but important question as this technique is favored by many vascular surgeons for asymptomatic lesions.
Go ahead. Unload the rest of your nitpicks here.
Thank you! Don’t mind if I do:
– The antiplatelet protocols were more aggressive in the CAS arm than that required for CEA.
– The design was intent-to-treat. Nearly 20% of the patients assigned to medical management eventually underwent a carotid revascularization.
– About two thirds of the patients enrolled had a carotid stenosis of 70-79%. My personal practice is not to perform routine CEA in asymptomatic lesions less than 80%.
So, what can we conclude from CREST-2?
Look, sequels can be difficult to pull off. For every Godfather II we get a dozen Deuce Bigalow European Gigolos. Still, CREST-2 was meticulously designed and implemented. It is now up to us to interpret these results in our real-world settings. We have data from thousands of patients enrolled in carotid revascularization trials. The CREST-2 results should enhance this abundance of knowledge, not replace it wholesale.
Clearly medical management is and remains the cornerstone of therapy. Get the LDL under 70, push the SBP below 130, stop smoking, and deploy a PCSK9 inhibitor when statins fail. For interventions, scrutinize your own outcomes. If your complication rates drift above trial level performance, you are not doing CREST-2, you are doing roulette. For every provider who interprets these results as an indication to routinely stent asymptomatic lesions, there is at least a 50% chance the CREST-2 investigators would not credential them.
No matter what technique you employ, the margin of benefit for carotid revascularization in asymptomatic lesions is razor thin. Following CREST-2, it does not seem like a leap to suggest that patient selection and provider skill matter as much or more than the type of intervention employed. How payers weaponize the CREST-2 findings remains to be seen. Blanket denials for asymptomatic revascularization is obviously a possibility. A better and more evidence-based approach would favor a requirement for registry-anchored proof of outcomes. CEA isn’t dead, but a strict diet is likely.
